Set ubatch sweep to requested 32k matrix

docs/LLM/ubatch-benchmark-large.md

@@ -22,10 +22,9 @@ for auditability only. Their results are invalid and must not be used.
 
 ## Sweep
 
-The default points are `256, 512, 768, 1024, 1280, 1536, 2048, 3072, 4096,
+The requested points are `256, 512, 768, 1024, 1280, 1536, 1792, 2048,
-5120, 6144`. This implements the requested 256-step range through 1536 and
+3072, 4096` at 32K context. Each point runs three 128-token repetitions by
-1024-step range from 2048 through 6144. Each point runs three 256-token decode
+default.
-requests by default.
 
 Run on Dankclaw from the repository checkout:
 
@@ -35,10 +34,10 @@ sudo systemctl stop frank-default-large
 sudo systemctl start frank-default-large
 ```
 
-The script uses the active Q6 AGENT model, a fixed 16K prompt workload, and
+The script uses the active Q6 AGENT model, a fixed 32K prompt workload, and
-`llama-bench` JSONL output. The initial ubatch-only sweep is:
+`llama-bench` JSONL output. The ubatch-only sweep is:
 ```text
-256, 512, 768, 1024, 1280, 1536, 1792, 2048, 3072, 4096, 5120, 6144
+256, 512, 768, 1024, 1280, 1536, 1792, 2048, 3072, 4096
 ```
 
 The GPU is shared with live workloads; do not run with the production service

scripts/bench-ubatch.sh

@@ -2,8 +2,9 @@
 # Benchmark ubatch with llama-bench for frank-default-large.
 #
 # This is intentionally a ubatch-only benchmark. It uses one llama-bench
-# process, a fixed 16K prompt workload, and a fixed batch size large enough to
+# process, a fixed 32K prompt workload, and a fixed scheduler size large
-# hold that workload. No server process, HTTP requests, or log scraping are used.
+# enough to hold that workload. No server process, HTTP requests, or log
+# scraping are used.
 #
 # Run on Dankclaw after stopping frank-default-large:
 #   sudo systemctl stop frank-default-large
@@ -13,8 +14,8 @@ set -euo pipefail
 
 : "${LLAMA_BENCH:=/home/dank/GitPackages/ROCmFPX/build-strix-rocmfp4/bin/llama-bench}"
 : "${MODEL:=/home/dank/models/kingjones-Qwen3.8-27B-STRIX-MTP/Qwen3.8-27B-Uncensored-Q6_0_ROCMFPX_AGENT.gguf}"
-: "${UBATCH_VALUES:=256,512,768,1024,1280,1536,1792,2048,3072,4096,5120,6144}"
+: "${UBATCH_VALUES:=256,512,768,1024,1280,1536,1792,2048,3072,4096}"
-: "${CONTEXT:=16384}"
+: "${CONTEXT:=32768}"
 : "${GEN_TOKENS:=128}"
 : "${REPETITIONS:=3}"
 : "${DEVICE:=ROCm0}"